Designing passive synchronization for MPI-2 one-sided communication to maximize overlap

Santhanaraman, Gopalakrishnan; Narravula, S.; Panda, D.K.

doi:10.1109/ipdps.2008.4536286

Cited by 9 publications

(4 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The RMA is a new MPI feature, requiring an external progress engine to remain truly passive-target [13], and not fully supported by all MPI libraries. Normally the MPI progress engine on each process activates only when the process itself is doing an active MPI communication.…”

Section: Implementation Of the Task-stealing Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

On a Dynamic Scheduling Algorithm for Massively Parallel Computations of Atomic Isotope

Dorofeeva¹,

Arora²,

Typel³

et al. 2016

Proceedings of the VII European Congress on Computational Methods in Applied Sciences and Engineering (ECCOMAS Congress 2016)

View full text Add to dashboard Cite

The goal of the work is to enhance the execution scalability of the code called AtomicClusters for evaluations of in-medium properties of nuclear clusters to extreme scales. In order to fully exploit the computing power of massively parallel supercomputing systems, the code was supplemented with parallel output and dynamic scheduling system based on task-stealing technique. The scheduling system was implemented for state-of-the-art distributed-memory high-performance computers (HPC) using the advanced features of Message Passing Interface (MPI) as an independent adjustable module. The parallel output was integrated into the code using MPI IO. A number of strong scaling tests was performed for the resulting parallel software. An almost linear scalability was reached on up to 4000 cores. The code scales up to at least 38400 processes, but with lower speedup. The obtained results are discussed in the fifth section of the paper.

show abstract

Section: Implementation Of the Task-stealing Algorithmmentioning

confidence: 99%

“…The ordered polling with match memorization was chosen as a balancing technique. The tests were run with all three MPI libraries installed on the system: MVAPICH [13], Intel MPI and Open MPI. The default settings of the libraries were used.…”

Section: Evaluation Of the Dynamic Scheduling Systemmentioning

confidence: 99%

On a Dynamic Scheduling Algorithm for Massively Parallel Computations of Atomic Isotope

Dorofeeva¹,

Arora²,

Typel³

et al. 2016

Proceedings of the VII European Congress on Computational Methods in Applied Sciences and Engineering (ECCOMAS Congress 2016)

View full text Add to dashboard Cite

show abstract

“…The authors in [51] have studied different optimization for reducing the synchronization overhead involved in implementing MPI one-sided communication. Designs for MPI RMA in InfiniBand clusters has been described in [52] and [53].In [54] and [55], the author describes a design for efficient passive synchronization using hardware support from InfiniBand atomic operations. In [56], authors discuss some performance guidelines for one-sided communication in MPI.…”

Section: Related Workmentioning

confidence: 99%

Runtime Support for Irregular Computation in MPI-Based Applications

Zhao

Balaji

Gropp

2015

2015 15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing

View full text Add to dashboard Cite

In recent years there are increasing number of applications that have been using irregular computation models in various domains, such as computational chemistry, bioinformatics, nuclear reactor simulation and social network analysis. Due to the irregular and datadependent communication patterns and sparse data structures involved in those applications, the traditional parallel programming model and runtime need to be carefully designed and implemented in order to accommodate the performance and scalability requirements of those irregular applications on large-scale systems. This thesis consists of two subtopics. The first subtopic focuses on improving MPI runtime to support the irregular applications from perspective of scalability and performance. The first three parts in this subtopic focus on MPI one-sided communication. In the first part, we present a thorough survey of current MPI one-sided implementations and illustrate scalability limitations in those implementations. In the second part, we propose a new design and implementation of MPI one-sided communication, called ScalaRMA, to effectively address those scalability limitations. The third part in this subtopic focuses on various issuing strategies in MPI one-sided communication. We propose an adaptive issuing strategy which can adaptively choose between delayed issuing strategy and eager issuing strategy in MPI ii runtime to achieve high performance based on current communication volume in MPI-based application. The last part in this subtopic is to tackle the scalability limitations in the virtual connection (VC) objects in MPI implementation. We propose a scalable design to reduce the memory consumption of VC objects in MPI runtime.The second subtopic of this thesis focuses on improving MPI programming model to better support the irregular applications. Traditional two-sided data movement model in MPI standard designed for scientific computation provides a paradigm for user to specify how to move the data between processes, however, it does not provide interface to flexibly manage the computation, which means user needs to explicitly manage where the computation should be performed. This model is not well suited for irregular applications which involve irregular and data-dependent communication pattern. In this work, we combine Active Messages (AM), an alternative programming paradigm which is more suitable for irregular computations, with traditional MPI data movement model, and propose a generalized MPI-interoperable Active Messages framework (MPI-AM). The framework allows MPI-based applications to incrementally use AMs only when necessary, avoiding rewriting the entire MPI-based application. Such framework integrates data movement and computation together in the programming model and MPI can coordinate the computation and communication in a much more flexible manner. In this subtopic, we propose several strategies including message streaming, buffer management and asynchronous processing, in order to efficiently handle AMs inside MPI. We al...

show abstract

“…MPICH2 [24] and LAM/MPI [2] designed the RMA communication on top of two-sided operations. MVAPICH2 has implemented a truly one-sided design [17] with special optimization on passive synchronization [18,23], but it only applies to the inter-node communication. In [22], a true one-sided passive synchronization scheme is designed for multi-core based systems, but it does not deal with active synchronization and data transfer operations.…”

Section: Related Workmentioning

confidence: 99%

Designing truly one-sided MPI-2 RMA intra-node communication on multi-core systems

2010

Self Cite

View full text Add to dashboard Cite

The increasing popularity of multi-core processors has made MPI intra-node communication, including the intra-node RMA (Remote Memory Access) communication, a critical component in high performance computing. MPI-2 RMA model includes one-sided data transfer and synchronization operations. Existing designs in popularly used MPI stacks do not provide truly one-sided intranode RMA communication. They are built on top of twosided send-receive operations, therefore suffering from overheads of two-sided communication and dependency on the remote side. In this paper, we enhance existing shared memory mechanisms to design truly one-sided synchronization. In addition, we design truly one-sided intra-node data transfer using two kernel based direct copy alternatives: basic kernel-assisted approach and I/OAT-assisted approach. Our new design eliminates the overhead of using two-sided operations and eliminates the involvement from the remote side. We also propose a series of benchmarks to evaluate various performance aspects over multi-core architectures (Intel Clovertown, Intel Nehalem and AMD Barcelona). The results show that the new design obtains up to 39% lower latency for small and medium messages and demonstrates 29% improvement in large message bandwidth. Moreover, it provides superior performance in terms of better scalability, reduced cache misses, higher resilience to process skew and increased computation and communication overlap. Finally, up to 10% performance benefits is demonstrated for a real scientific application AWM-Olsen.

show abstract

Designing passive synchronization for MPI-2 one-sided communication to maximize overlap

Cited by 9 publications

References 13 publications

On a Dynamic Scheduling Algorithm for Massively Parallel Computations of Atomic Isotope

On a Dynamic Scheduling Algorithm for Massively Parallel Computations of Atomic Isotope

Runtime Support for Irregular Computation in MPI-Based Applications

Designing truly one-sided MPI-2 RMA intra-node communication on multi-core systems

Contact Info

Product

Resources

About